Measuring the effect of historical structural racism on community firearm violence in US cities

INTRODUCTION

Community firearm violence in the United States (US), defined as fatal and nonfatal interpersonal shootings in public spaces,¹ disproportionately affects neighborhoods where people racialized as Black predominately live. While previous research has established structural racism as a fundamental cause of community firearm violence,^2,3 methods to measure and quantify structural racism vary widely. Early attempts to measure the impact of structural racism on community firearm violence were limited to isolated aspects of this multifaceted issue, like residential segregation^2,4 or economic inequality,^5,6 both of which were found to correspond with increases in firearm violence. However, structural racism is the result of mutually reinforcing factors each playing a role in perpetuating inequity.⁷

In recent years, research examining the association between structural racism and community firearm violence has broadened to encompass multiple domains when measuring structural racism.^8,9 Researchers have created indices/scales and factor scores to explore the intricate interactions within structural racism.³ Despite these advancements, there is a gap in how this field approaches the subject. Current studies rely on a single historical period to examine the relationship between structural racism and present-day community firearm violence. Incorporating a historical perspective is essential to fully comprehend how past policies and societal changes have shaped current patterns of structural racism, providing a more accurate understanding of its long-term effects on community firearm violence.

Critical race theory may aid in identifying the historical determinants of community firearm violence. An academic and legal framework, critical race theory posits racism as not merely the product of individual bias or prejudice, but a more deeply ingrained, enduring, and central component of society and its institutions.¹⁰ By examining the historical context of structural racism, we can more accurately trace the origins and evolution of policies, understand their short- and long-term effects, and recognize enduring patterns that continue to shape present-day community firearm violence.

In the early 1900s, racial residential segregation in the US exposed the deeply ingrained discriminatory policies and institutional practices that shaped the urban landscape. Before the Fair Housing Act of 1968,¹¹ segregation-based policies and practices created racial divisions within US cities. Racist practices of the Home Owners’ Loan Corporation (HOLC) in the 1930s,¹² where “undesirable” neighborhoods were graded with a D and labeled “red” have become the de jure legacy of modern racial segregation and home ownership through desegregation. This period also saw the establishment of sundown towns, where non-White individuals were forced to leave before sunset to avoid harm,¹³ and the widespread use of lynchings to violently enforce racial boundaries.¹⁴ These state-sanctioned acts of violence, often inflicted by police and residents, left many individuals racialized as Black without justice. Together, these three phenomena—redlining, sundown towns, and lynchings—represent extreme exposure to racism that solidified residential segregation in US cities before the Fair Housing Act of 1968, which amended the Civil Rights Act of 1964.

The Civil Rights Movement marked significant progress toward desegregation by dismantling overtly racist laws and practices across the US. However, by 1990, structural racism had evolved, becoming more fragmented and manifesting as institutional racism within specific sectors such as criminal justice, education, housing, and employment. This shift meant that while overt segregation was legally challenged, discriminatory practices persisted in more subtle and systemic forms within individual institutions. These ongoing inequities continued to affect marginalized communities in diverse and pervasive ways, demonstrating that the struggle against racism had shifted from overt segregation to more insidious forms of institutional discrimination.

Modern-day practices have shifted to “color-blind” discrimination, where policies and actions seemingly do not target individuals racialized as Black, yet disproportionately impact them. Ostensibly neutral policies, particularly in housing and employment, perpetuate systemic disparities rooted in historical and ongoing biases. This covert form of racism operates under the guise of equality but continues to sustain and even exacerbate the inequities that the Civil Rights Movement sought to eliminate. Examining the enduring impact of structural racism reveals how historical policies still shape present-day challenges, including disparities in exposure to community firearm violence. Communities with high concentrations of minoritized residents continue to face residential segregation, lack of economic investment, limited access to essential resources, elevated pollution levels, and vulnerability to gentrification.^15–17 Critical race theory, with its emphasis on the experiential knowledge of marginalized groups, highlights the need to acknowledge these historical and contemporary forms of structural racism in the study of firearm injury epidemiology.

Our study examines the effect of structural racism over time on contemporary patterns of community firearm violence in US cities. We focus on cities due to the distinct dynamics of firearm violence across the urban-rural continuum.¹⁸ Research indicates that while firearm violence is a nationwide issue, its manifestations vary considerably between different regions. In rural counties, firearm suicide tends to be more prevalent, whereas urban counties experience higher rates of firearm homicide.¹⁸ Additionally, the disproportionate impact of firearm mortality on individuals racialized as Black is most pronounced in urban settings, where they constitute 68% of homicide victims.¹⁹ Due to data limitations, we examine three distinct historical periods to assess whether the legacy of structural racism has increasingly influenced community firearm violence in the present day. We hypothesize that the cumulative effects of structural racism will result in a stronger association between the most recent period and contemporary firearm violence. We further hypothesize that modern discriminatory policies and practices, as evolved forms of past injustices, have had a more pronounced impact on current patterns of community firearm violence.

METHODS

Study sample

This study utilized the list of 500 cities identified by the Centers for Disease Control and Prevention’s (CDC) 500 Cities Project (currently known as PLACES) to define its sample. The 500 Cities Project selected the 500 largest US cities based on population size according to the US 2010 Census. Information on the 500 cities, as well as the complete description of the dataset, is publicly available and can be downloaded from the CDC website.²¹

Variables

Exposure variable.

We explored the enduring impact of historical structural racism on city-level community firearm violence across three historical periods, adapted from Adkins-Jackson and colleagues’ model of life course socioeconomic racism:²² Historical Period 1-Before the Fair Housing Act of 1968 (1900–1968), Historical Period 2-After Desegregation (1970–2000), and Historical Period 3-Modern Times (2010–19) (Figure 1). For each period, variable selection was guided by a theoretical understanding of structural racism, building upon the life course model of socioeconomic racism. The focus was on variables that capture population-based Black-non-Black disparities resulting from racialization and discrimination processes occurring between 1900 and 2019.

Figure 1.

Full conceptual model demonstrating the impact of historical structural racism on community firearm violence in cities

Historical Period 1.

In Historical Period 1 (1900–1968) racist institutional practices such as redlining, the enforcement of sundown towns, and state-sanctioned lynchings were instrumental in creating and maintaining racial residential segregation.¹⁴ To operationalize structural racism during this period, we created a binary variable to indicate whether a city had an existing HOLC map, was identified as a sundown town, or had one or more lynchings of an individual racialized as Black. We coded cities meeting any one of these criteria as having a “documented history” of racism as 1 and cities without documentation as 0.

The decision to include these three variables for Historical Period 1 was influenced by several key factors. Primarily, we were constrained by the availability of data for this historical period. We relied on data from three major sources: Mapping Inequality,²³ History and Social Justice,¹³ and Equal Justice Initiative.²⁴ The Mapping Inequality project provides comprehensive maps and documents detailing the historical redlining practices of the HOLC. The History and Social Justice Initiative documents and analyzes the historical and ongoing impacts of racial discrimination and social injustice in the US and houses the only registry of sundown towns. The Equal Justice Initiative, a nonprofit organization dedicated to ending mass incarceration, offers the most complete documentation of lynchings of individuals racialized as Black.

Second, the complexity and variability of these measures across different cities posed challenges in creating a standardized, comparable framework. Our goal was to utilize the most direct and consistently available indicators to reliably reflect broader patterns of structural racism, particularly as they manifested through racial residential segregation across various urban settings. Additionally, we chose the term “documentation” to emphasize that a city might appear non-racist simply due to a lack of recorded evidence, reflecting the limitations in the current landscape of quantifiable data sources on structural racism. Data were available at the city level for each variable, except for lynching data, where we used the centroid of each city to nest cities into counties, relying on county-level data as the smallest geographic unit of analysis available.

Historical Period 2.

To assess the impact of structural racism during Historical Period 2 (1970–2000), we examined several indicators: home ownership via owner-occupied units, unemployment rates, the proportion of individuals living below the federal poverty line, jail population, and educational attainment measured by the lack of a high school diploma. For this analysis, we utilized data from the IPUMS Census²⁵ and the Vera Institute of Justice.²⁶ The IPUMS Census provides decennial censuses from 1790 to 2010 and American Community Surveys (ACS) from 2000 onward. The Vera Institute of Justice offers comprehensive data on incarceration rates, trends, and disparities in the US.

We calculated the proportion of individuals racialized as Black experiencing each variable by dividing the total number of people racialized as Black who owned a unit they occupied, were unemployed, lived below the poverty line, were incarcerated in a jail, or did not have a high school diploma by the total number of people for each respective variable, per city in 1990. With the exception of jail incarceration data, which followed the same methodology as the lynching data, all information was available at the city level. We then dichotomized all proportions into two categories: highest racism (coded as 1), using the top quartile as the cutoff, and all other values (coded as 0).

We then conducted a confirmatory factor analysis (CFA) within an Oblimin rotation and diagonally least-weighted squares estimator with all variables at Historical Period 2. Model fit statistics for the CFAs include both comparative fit index (CFI) and Tucker-Lewis index (TLI) between 0·90 and 1, and root mean square error of approximation (RMSEA) <0·10.²⁷ Where available robust fit statistics were recorded. Factor loadings for variables in the CFA were observed (>0·5). Results yielded one continuous score (−1 to 1) per city. Finally, we coded values over the top quartile (1=evidence of high structural racism, 0=evidence of low-mid structural racism).

Historical Period 3.

For Historical Period 3 (2010–19) we operationalized this period of “color blind” structural racism with mortgage lending denial, police killings, arrests, water pollution, food deserts, and per pupil revenue. We used data from the Home Mortgage Disclosure Act,²⁸ Environmental Protective Agency,²⁹ US Department of Agriculture,³⁰ Mapping Police Violence,³¹ Federal Bureau of Investigation’s Uniform Crime Reports,³² and National Neighborhood Data Archive.³³ The Home Mortgage Disclosure Act provides comprehensive data on home loan applications and denials, offering insights into lending discrimination. The Environmental Protection Agency offers detailed data on water pollution, including contaminants, sources, and regulatory compliance. The U.S. Department of Agriculture identifies food deserts, highlighting areas with limited access to affordable, nutritious food. Mapping Police Violence supplies extensive data on police killings, detailing the demographics of victims, incident circumstances, and racial disparities in police use of force. The Uniform Crime Reports compile nationwide crime data, including statistics on violent crimes, property crimes, arrests, and related trends. Lastly, the National Neighborhood Data Archive provides geographically referenced data on neighborhood characteristics, including per-pupil revenue, enabling analysis of educational funding disparities.

For mortgage lending denial, police killings, and arrests, we computed proportions following the same method as Historical Period 2. For variables where race was not explicitly part of the raw data (i.e., water pollution, food deserts, and per pupil revenue), we identified cities with one or more health-based drinking water violations, a majority of census tracts classified as food insecure, and school districts falling into the bottom quartile of per pupil revenue. We then classified cities as having evidence of structural racism (coded as 1) if they also had a proportion of residents or students (school districts) racialized as Black that was equal to or greater than the national average; all other cities were coded as 0. Data for water pollution and police killings were available at the city level, while the other variables were treated similarly to those outlined for previous periods. Finally, we conducted a CFA for all variables in Historical Period 3, following the same methods used for Historical Period 2. The resulting factor scores were then dichotomized, with the top quartile representing evidence of high structural racism (coded as 1) and all other scores indicating low-to-mid structural racism (coded as 0).

Outcome variable.

We used data from the Gun Violence Archive (GVA), an independent data collection and research group gathering daily information on firearm violence incidents from public records.³⁴ Our study focused on interpersonal shootings to examine the relationship between structural racism and community firearm violence. Including firearm suicides would introduce different dynamics, as factors influencing suicides differ significantly from those affecting interpersonal shootings. Additionally, the GVA does not include detailed information on firearm suicides. Available fields in the GVA data describe the date when the shooting occurred, the address, geographic coordinates of the shooting event, and the total number of individuals killed or injured in the incident. These data are publicly available through the GVA website and searchable starting from 2013. We geocoded and aggregated the data across the 5-year period (2015–19) to calculate the total number of all shootings, fatal shootings, and nonfatal shootings for each city. We then calculated the rate of shootings (all, fatal, and nonfatal) per 100,000 population using city population estimates from the American Community Survey (2015–19). Finally, we log-transformed the outcome due to significant right-skewness in the data. We applied a log base 2 scale to simplify the interpretation of our findings.

Data Analyses

We began by assessing the correlations between the individual variables across all three time periods. Next, we fit independent semi-log-linear models to evaluate the relationship between each historical period and firearm violence outcomes (all, fatal, and nonfatal shootings) separately. Following this, we used a structural equation model (SEM) to examine the combined effects of all historical periods. Within the SEM framework, we employed base 2 semi-log-linear regression models to evaluate the relationships between the three historical periods and the firearm violence outcomes collectively.

RESULTS

Out of the initial 500 cities, 397 were selected for inclusion in the final sample as they had complete data available for all three historical periods (Table 1). Included cities collectively encompassed 239 counties across 40 different states. From 2015 to 2019, there were 96,028 total shootings in the included cities, 70% of which were nonfatal. Included cities were statistically significantly similar to excluded cities for all variables except arrests in Historical Period 3 where the average proportion of individuals racialized as Black arrested was 25·4% in included cities compared to 30·3% in excluded cities (p-value = 0·02). The results from the correlation matrix can be seen in Table 2.

Table 1.

Descriptive characteristics of US cities included in the study (n=397)

Variable	# of Cities (%)	Mean (SD)
All shootings (per 100,000)		964·7 (128·0)
Fatal shootings (per 100,000)		314·1 (36·3)
Nonfatal shootings (per 100,000)		650·6 (95·3)
Historical Period 1: Before the Fair Housing Act of 1968 (1900–68)
Redlining	114 (28·7)
Sundown town	78 (19·6)
Lynchings	119 (30·0)
Historical Period 2: Desegregation (19702000)
Unemployment (%)		19·9 (20·9)
Jail (%)		36·1 (22·6)
Owner occupied units (%)		8·6 (11·9)
Poverty (%)		22·0 (22·5)
No high school diploma (%)		14·9 (18·2)
Historical Period 3: Modern Times (2010–19)
Mortgage lending denial (%)		11·8 (13·7)
Police killings (%)		27·1 (33·3)
Arrests (%)		25·5 (18·9)
Water pollution	96 (24·2)
Food desert	80 (20·2)
Per pupil revenue	241 (60·7)

Table 2.

Correlation matrix for individual variables in each time period

	HP1: Redline/Sundown Town/Lynching	HP2: Educational Attainment	HP2: Unemployment	HP2: Jail Population	HP2: Home Ownership	HP2: Poverty	HP3: Mortgage Lending Denial	HP3: Police Killings	HP3: Arrests	HP3: Water Pollution	HP3: Food Deserts	HP3: Per Pupil Revenue
HP 1: Redline/Sundown Town/Lynching	1	0.498	0.453	0.459	0.348	0.466	0.387	0.373	0.456	0.334	0.201	0.466
HP2: Educational Attainment		1	0.979	0.802	0.936	0.972	0.882	0.689	0.743	0.662	0.591	0.788
HP2: Unemployment			1	0.776	0.955	0.986	0.904	0.741	0.762	0.699	0.599	0.783
HP2: Jail Population				1	0.690	0.794	0.717	0.662	0.838	0.443	0.283	0.830
HP2: Home Ownership					1	0.925	0.895	0.690	0.678	0.605	0.577	0.659
HP2: Poverty						1	0.889	0.687	0.765	0.686	0.580	0.792
HP3: Mortgage Lending Denial							1	0.654	0.744	0.577	0.559	0.740
HP3: Police Killings								1	0.622	0.381	0.323	0.581
HP3: Arrests									1	0.407	0.501	0.859
HP3: Water Pollution										1	0.707	0.729
HP3: Food Deserts											1	0.708
HP3: Per Pupil Revenue												1

HP1: Historical Period 1

H2: Historical Period 2

HP3: Historical Period 3

Confirmatory Factor Analyses (CFAs)

For Historical Period 2, a 1-factor model achieved adequate model fit based on a CFI of 0·996, TLI of 0·992, and RMSEA of 0·061. Unemployment loaded highest (0·966) and jail population loaded lowest (0·590). Adequate fit and high factor loadings suggest these variables encompass a unidimensional latent variable (see Table 2 for fit statistics). Similarly, for Historical Period 3, a 1-factor model achieved a modest fit (CFI 0·902, TLI 0·836, RMSEA 0·124). Mortgage lending denial loaded highest (0·703) and police killings loaded lowest (0·522). Despite challenges, these values suggest these variables encompass a unidimensional latent variable (see Table 3 for fit statistics).

Table 3.

Factor loadings and fit statistics

Historical Period 2- After Desegregation (1970–2000)		Historical Period 3- Modern Times (2010–19)
Variable	Loading	Variable	Loading
Unemployment	0·966	Mortgage lending denial	0·703
Jail	0·590	Water pollution	0·534
Owner occupied units	0·824	Food desert	0·522
Poverty	0·924	Police killings	0·541
No high school diploma	0·918	Arrests	0·663
		Per pupil revenue	0·660
Fit statistics		Fit statistics
CFI	0·996	CFI	0·902
TLI	0·992	TLI	0·836
RMSEA	0·061	RMSEA	0·124

Independent associations

Structural racism in each historical period was independently and significantly associated with an increased risk of city-level firearm violence (Table 4). Cities with a documented history of structural racism during Historical Period 1 had a higher rate of shootings per 100,000 people compared to cities without documented history of structural racism, with a point estimate of 0.84 (95% CI: 0.50, 1.18) on the log2 scale. When exponentiated, this corresponded to a 1.79 fold increase, or a 79% higher rate of all shootings per 100,000 people (95% CI: 1.41, 2.27), compared to cities without documented evidence of structural racism. Additionally, cities with evidence of high levels of structural racism were associated with a higher rate of shootings per 100,000 people in Historical Period 2 (beta: 2.26, 95% CI: 1.94, 2.58) and Historical Period 3 (beta: 2.31, 95% CI: 1.99, 2.63) on the log2 scale, compared to cities with evidence of low-mid levels of structural racism. We observed similar results when disaggregating fatal and nonfatal shootings. The strongest effects were among nonfatal shootings.

Table 4.

Semi-log-linear¹ model results estimating the association between individual historical periods and community firearm violence in US cities, 2015–19 (n=397)

	log2(All Shootings per 100K)		log2(Fatal Shootings per 100K)		log2(Nonfatal Shootings per 100K)
	Beta	95% CI	Beta	95% CI	Beta	95% CI
Historical Period 1 - Before the Fair Housing Act of 1968 (1900–68)	0.84	(0.50, 1.18)	0.55	(0.25, 0.85)	1.00	(0.61, 1.40)
Historical Period 2 - After Desegregation (1970–2000)	2.26	(1.94, 2.58)	1.90	(1.62, 2.19)	2.52	(2.14, 2.90)
Historical Period 3 - Modern Times (2010–19)	2.31	(1.99, 2.63)	1.86	(1.56, 2.15)	2.62	(2.25, 3.00)

¹The outcome variable is log transformed using log base 2 scale

Bolded results are statistically significant at an alpha of 0.05

Structural Equation Model (SEM)

All historical periods had similar associations as the independent model results with documented/high structural racism associated with greater city-level firearm homicides (Figure 2). A table of all results (i.e., all, fatal, and nonfatal shootings) can be found in Appendix A. Cities with a documented history of structural racism during Historical Period 1 had a higher rate of shootings per 100,000 people compared to cities without documented history of structural racism, with a point estimate of 0.34 (95% CI: 0.06, 0.62) on the log2 scale. When exponentiated, this corresponded to a 1.27 fold increase, or a 27% higher rate of all shootings per 100,000 people (95% CI: 1.04, 1.54), compared to cities without documented evidence of structural racism. Additionally, cities with evidence of high levels of structural racism were associated with a higher rate of shootings per 100,000 people in Historical Period 2 (beta: 1.22, 95% CI: 0.78, 1.65) and Historical Period 3 (beta: 1.35, 95% CI: 0.91, 1.79) on the log2 scale, compared to cities with evidence of low-mid levels of structural racism. A documented history of structural racism at Historical Period 1 predicted evidence of high structural racism at Historical Period 2 (beta: 0·23, 95% CI: 0·14, 0·31), and evidence of high structural racism at Historical Period 2 predicted evidence of high structural racism at Historical Period 3 (beta: 0·71, 95% CI: 0·64, 0·77) on the log2 scale.

Figure 2.

Structural equation model semi-log-linear results for the association between historical structural racism and all shootings in cities, 2015–19 (n=397)

DISCUSSION

This study examined the effect of historical structural racism on contemporary patterns in community firearm violence in US cities. This study explored this association across three distinct historical periods of structural racism with a primary focus on uncovering the persistent effects of structural racism on the risk of community firearm violence. The results present robust and compelling evidence of the enduring influence of structural racism on community firearm violence in cities. Notably, we observed significant associations between each of the historical periods and an increased risk of community firearm violence. This particularly holds true for Historical Period 3 affirming our hypothesis that there will be a stronger association between the most recent period of structural racism and community firearm violence.

Our findings echo the insights from previous research which find that structural racism is a fundamental cause of firearm violence.^2,3 Many of these studies, however, have traditionally focused on specific facets of structural racism, such as residential segregation and redlining, which contributed to concentrated poverty, limited economic opportunities, and, subsequently, higher rates of firearm violence.^2,4–6 Recent research highlights the importance of examining structural racism across multiple domains rather than isolating individual components. In our study, we found that racial segregation and economic opportunity are particularly critical factors in understanding the path from structural racism to community firearm violence. These findings align with previous work on the association between income inequality and violent crimes, including firearm-related homicides in the US.^6,35–37 Our study contributes to this literature by highlighting the importance of integrating a temporal perspective into the study of structural racism.

Associations between the measures of structural racism and community firearm violence were strong and increased in magnitude over time. These results are consistent with our theoretical perspective that multiple domains of structural racism compound over time to shape the health of populations, including experience of firearm violence. When structural racism is operationalized through a range of variables—such as housing discrimination, unequal access to quality education, policing practices, and economic exclusion—it becomes evident how these interconnected systems can lead to extreme outcomes, including heightened rates of firearm violence. By assessing structural racism across these diverse domains, our study captures the multifaceted and cumulative impacts that are often missed when examining individual factors in isolation. This comprehensive approach emphasizes how deeply embedded and pervasive these inequities are, ultimately driving the severe disparities observed in community violence.

Guided by critical race theory³⁸, a central finding from this research is the persistent effect of structural racism over time. Critical race theory calls for a centering, or de-marginalization, of racialized and minoritized people in the world. In the context of the US, this includes people racialized as Black who through science and other forms of storytelling have described the effects of structural racism on the community violence they experience. From Billie Holiday’s 1939 “Strange Fruit”, Nina Simone’s 1964 “Mississippi Goddam”, Ice Cube’s 1993 “Ghetto Bird”, to Lil Baby’s 2020 “The Bigger Picture”—people racialized as Black have repeatedly named structural racism as the fundamental cause of violence. This study quantitatively validates those lived experiences as applied to the specific problem of firearm violence.

Moreover, our historical methodology offers a valuable framework for exploring the structural factors influencing firearm deaths beyond urban environments and racial dynamics. While our analysis primarily focuses on urban firearm homicides linked to systemic inequities among individuals racialized as Black, this approach can also be applied to examine the distinct structural influences driving other types of firearm violence. For example, suicides often stem from economic and social pressures, including the effects of national and international policies, such as those governing agriculture, which have reshaped rural economies and exacerbated social isolation.³⁹ Although these policies may not be directly motivated by racial bias, they have nonetheless contributed to a landscape of despair and economic instability, leading to increased suicide rates.^40,41 Applying our methodology to these contexts could reveal the complex interactions between economic conditions, cultural attitudes, and firearm violence across different regions and populations, potentially informing more comprehensive policy responses to address the underlying causes of firearm deaths in various communities.

The findings from our research extend beyond firearm epidemiology. For example, the COVID-19 pandemic highlighted racial health disparities, with individuals racialized as Black experiencing significantly higher exposure and mortality rates compared to White indivdiuals.⁴² However, the connections between different facets of structural racism can be complex, and their unique and joint effects on health are not always immediately evident. By deconstructing structural racism across historical periods, our research offers a framework that can be applied to understand the experiences of other racial and ethnic groups facing discrimination, both within and beyond the US.

Limitations

In general, it is challenging for quantitative analyses of extended historical processes to fulfill the prerequisites for accurately pinpointing unbiased causal effects. A significant constraint in this context arises from the scarcity of historical data that could be employed to fulfill requirements related to conditional exchangeability assumptions. Despite advancements in data curation, allowing researchers to partially reconstruct the social contexts influencing the distribution of historical treatments and their subsequent outcomes, the lack of widespread, standardized data collection mechanisms hampers the capacity to consider historical information and ensure confidence that unobserved factors do not influence outcomes. As such, we frame the results as having evidence of high-level structural racism and evidence of low to mid-levels of structural racism. However, we emphasize that evidence does not negate the presence of structural racism.

Further data availability limitations exist regarding both the exposure and outcome data. The original sample of 500 cities was reduced to 397 cities based on data availability across all three historical periods. This reduction in the sample size raises concerns about selection bias as the omitted cities may possess distinct characteristics or experiences that differ from those included in the final sample. However, included cities were not statistically significantly different from included cities with respect to all variables tested except for arrests in Historical Period 3. Additionally, while the theoretical framework acknowledges the potential value of extending the analysis to even earlier historical periods, the lack of comprehensive data constrains the ability to explore this possibility fully. Concerning outcome data, no comprehensive national data source for community firearm violence currently exists. Publicly available data from government agencies are typically only available for fatal injuries at the city or county level.⁴³ Data aggregated at this level may be too spatially coarse to be informative when studying the population impact of firearm violence at the neighborhood level.

Finally, a constraint lies in the potential for the latent construct of structural racism to limit implications for interventions that target one aspect of structural racism. This discord between demonstrating aggregate impact and practical intervention needs further implementation science research to bridge the gap to strategically combat structural racism and its associated impacts on firearm violence.

Conclusion

Our findings illuminate the long-term influence of historical structural racism on contemporary community firearm violence in a large number of US cities. Intervention strategies should account for the interconnectedness of structural racism across different historical periods and systems of inequality. A holistic approach that is both comprehensive and substantial in scope is needed to effectively reduce structural racism and promote community well-being. It is imperative that future efforts center on dismantling structural racism to achieve lasting social justice and community safety.

Highlights.

• Community firearm violence impacts marginalized groups in the US disproportionately

• Structural racism is a key factor driving disparities in rates

• Cities with a history of structural racism have higher shooting rates from 2015–19

• Need for research and interventions to address historical structural racism

Appendix

Appendix Table A.1.

Structural equation model semi-log-linear¹ results for the association between historical structural racism and community firearm violence in cities, 2015–19 (n=397)

	log2(All Shootings per 100K)		log2(Fatal Shootings per 100K)		log2(Nonfatal Shootings per 100K)
	Beta	95% CI	Beta	95% CI	Beta	95% CI
Historical Period 1 - Before the Fair Housing Act of 1968 (1900–68) on Historical Period 2 - After Desegregation (1970–2000)	0.23	(0.14, 0.31)	0.23	(0.14, 0.31)	0.23	(0.14, 0.31)
Historical Period 2 - After Desegregation (1970–2000) on Historical Period 3- Modern Times (2010–19)	0.71	(0.64, 0.77)	0.71	(0.64, 0.77)	0.71	(0.64, 0.77)
Historical Period 1 - Before the Fair Housing Act of 1968 (1900–68) on Shootings	0.34	(0.06, 0.62)	0.13	(−0.13, 0.38)	0.45	(0.12, 0.78)
Historical Period 2 - After Desegregation (1970–2000) on Shootings	1.22	(0.78, 1.65)	1.18	(0.79, 1.58)	1.27	(0.76, 1.78)
Historical Period 3 - Modern Times (2010–19) on Shootings	1.35	(0.91, 1.79)	0.97	(0.57, 1.37)	1.60	(1.08, 2.11)

¹The outcome variable is log transformed using log base 2 scale

Bolded results are statistically significant at an alpha of 0.05